Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
60 Years After Brown: Trends and Consequences of School Segregation
Sean F. Reardon Stanford University
Ann Owens
University of Southern California
Version: October 1, 2013
Sean F. Reardon Graduate School of Education Stanford University 520 Galvez Mall, #526 Stanford, CA 94305-3084 [email protected] (650) 736-8517 Ann Owens Department of Sociology University of Southern California 851 Downey Way, HSH 204 Los Angeles, CA 90089 [email protected] (213) 821-5730 Keywords: School Segregation; Race; Income
60 Years After Brown: Trends and Consequences of School Segregation
Abstract
Since the Supreme Court’s 1954 Brown v. Board of Education decision, researchers and
policymakers have paid close attention to trends in school segregation. Here we review the
evidence regarding these trends and their consequences. In general, the evidence regarding trends
in segregation suggests that the most significant declines in school segregation occurred at the end
of the 1960s and the start of the 1970s. Although there is disagreement about the direction of more
recent trends in segregation, this disagreement is largely driven by different definitions of
segregation and different ways of measuring it. We conclude that the changes in segregation in the
last few decades are not large, regardless of what measure is used, though there are important
differences in the trends across regions, racial groups, and institutional levels. In addition, we
discuss the role of desegregation litigation, demographic changes, and residential segregation in
shaping these trends.
One of the reasons that scholars, policymakers, and citizens are concerned with school
segregation is that segregation is hypothesized to exacerbate racial or socioeconomic disparities in
educational success. The mechanisms that would link segregation to disparate outcomes have not
often been spelled out clearly or tested explicitly. We develop a general conceptual model of how
and why school segregation might affect students and review the relatively thin body of empirical
evidence that explicitly assesses the consequences of school segregation. This literature suggests
that desegregation in the 1960s and 1970s was beneficial to blacks; evidence of the effects of
segregation in more recent decades, however, is mixed or inconclusive. We conclude with
discussion of aspects of school segregation on which further research is needed.
1
60 Years After Brown: Trends and Consequences of School Segregation
1. Introduction
In the 60 years since the Supreme Court’s 1954 Brown v. Board of Education decision (347
U.S. 483) outlawing de jure school segregation in American public schools, the patterns of
residential and school segregation in the United States have changed dramatically. These changes in
segregation patterns, however, have been inconsistent across time and place in both their pace and
direction. Prior to Brown, school segregation was absolute in the South, and very high in many
school districts in other parts of the country. Several forces have altered these patterns over the last
six decades, including continuing changes in the legal and policy landscape, demographic changes,
changes in residential segregation patterns, and changes in public attitudes regarding the value and
feasibility of school integration.
In this article, we review the evidence regarding these trends and their consequences. In
particular, we begin with an extensive review of the empirical research describing trends in school
segregation in the six decades since Brown. Because these trends differ depending on the type of
segregation (black-white, Hispanic-white, multiracial, or socioeconomic, for example) and the level
of aggregation (national, metropolitan, district, or school-level) of interest, there is no single answer
to the question of how school segregation has changed over the last 60 years. Moreover,
segregation can be measured in a number of ways, which further complicates simple descriptions of
segregation trends and patterns. Our goal in this first section of the article, then, is to provide a
review of the evidence on segregation trends and patterns across these multiple dimensions.
Second, we discuss the causes of the trends in school segregation. As we note, segregation
patterns have changed for a number of reasons in the last 60 years. A number of Supreme Court
decisions have changed the legal landscape of desegregation efforts. Demographic changes,
particularly the rapid growth of the Hispanic population, have changed the composition of the
2
school-age population. Declining residential racial segregation and rising income segregation have
changed the spatial distribution of families and patterns of school segregation over the last 50 years
(Jargowsky 1996; Charles 2003; Logan et al. 2004; Watson 2009; Logan & Stults 2011; Reardon &
Bischoff 2011a; 2011b; Glaeser & Vigdor 2012; Iceland & Sharp 2013). Finally, public opinion has
become more racially tolerant, but increasingly opposed to busing and other school segregation
practices (Orfield 1995).
Third, we review the evidence regarding the consequences of school segregation for
students. One of the reasons that many scholars, policymakers, and citizens are concerned with
school segregation is that segregation is hypothesized to exacerbate racial or socioeconomic
disparities in educational success. Our review of the literature, however, suggests that the
mechanisms that would link segregation to disparate outcomes have not often been spelled out
clearly or tested explicitly. Indeed, much of the research purporting to assess the links between
segregation and student outcomes tests instead measures the association between school
composition and student outcomes. Such research can be considered a test of the effects of
segregation only in a limited sense, under the assumption that segregation affects student outcomes
primarily through school composition mechanisms, rather than through other possible mechanisms
such as the unequal distribution of resources and disparities in school and teacher quality. Thus, we
begin our review of the literature on the consequences of segregation with a brief discussion of a
general conceptual model of how and why school segregation might affect students. Following this,
we review the relatively thin body of empirical evidence that explicitly assesses the consequences
of school segregation.
Despite the extensive body of research on trends and patterns of school segregation, and
the somewhat thinner body of research on its effects, a number of questions remain. We conclude
with discussion of questions where further research would be valuable.
3
2. Trends in School Segregation
Trends in school segregation may differ depending on the groups of interest (racial/ethnic
or socioeconomic groups) and the geographic scale and organizational units of interest (schools,
districts, metropolitan areas, and the nation). Most segregation research in the United States has
focused on black-white segregation between schools and within school districts. In part, the black-
white focus is driven by the historical legacy of slavery and the continuing salience of black-white
inequality; the within-district, between-school focus is driven by the fact that legal, policy, and
practical constraints make it easier to affect between-school segregation within districts than
segregation at larger (between-district) or smaller (within-school) institutional levels. Nonetheless,
any complete accounting of segregation patterns and trends must take into account segregation
among other racial/ethnic groups (including Hispanic-white segregation) and socioeconomic
segregation patterns, as well as between-district segregation. We review segregation trends along
each of these dimensions, to the extent there is available research, below. First, however, we
digress briefly to discuss the measurement of segregation.
2.a. Measures of Segregation
School segregation is typically measured using one of two types of segregation indices:
measures of isolation or exposure and measures of unevenness (Massey and Denton, 1988). These
different ways of measuring segregation often yield very different conclusions about the direction
and magnitude of trends in segregation.
Indices of unevenness measure the extent to which a student population is unevenly
distributed among schools. For example, the black-white dissimilarity index represents the
proportion of the black (or white) population who would have to change schools in order to yield a
pattern of school enrollment in which each school has identical racial proportions (Duncan &
Duncan 1955; James & Taeuber 1985; Massey & Denton 1988). Other indices of evenness include
Theil’s information theory index, the variance ratio index, and the Gini index of segregation (James
4
& Taeuber 1985; Massey & Denton 1988). These measures generally are scaled from 0 to 1, with 0
indicating no segregation (every school has the same racial composition) and 1 indicating complete
segregation (no child attends school with any other child of a different race); values above 0.60 are
considered indicative of “high segregation” (Massey & Denton 1989).
Indices of exposure or isolation, however, measure the extent to which students are
enrolled in schools with high or low proportions of a given racial group. For example, the black
isolation index is defined as the average proportion of black students in black students’ schools;
likewise, the white-black exposure index is the average proportion of black students in white
students’ schools (Coleman et al 1975; Lieberson & Carter 1982, Massey & Denton 1988). Another
measure of isolation that is sometimes used is the proportion of students who attend “racially
isolated” schools—often defined as schools with a high proportion of minority students (see, e.g.,
Orfield, 2001). Massey & Denton (1989) describe isolation indices above 0.70 (or, equivalently,
exposure indices below 0.30) as indicating “high segregation.”
The evenness measures and the exposure/isolation measures capture different dimensions
of segregation. To see this, consider a school district in which 90% of students are black. If all
schools in the district had enrollments that were 90% black, we would have low unevenness, but
high black isolation (or, equivalently, low black-white exposure), because the average black student
would attend a predominantly black school. Conversely, in a school district with very few black
students, isolation might be low even if students were very unevenly distributed by race. Put
differently, exposure and isolation measures are sensitive to the overall racial composition of a
school district, while the evenness measures are not.
This distinction has implications for any assessment of trends in segregation, because
changing racial population composition may lead to increases in measured isolation, even if the
extent to which students are evenly/unevenly distributed among schools doesn’t change. However,
there is not one “correct” measure of segregation. To the extent that we think that segregation
5
affects students through peer or compositional effects, or mechanisms correlated with school
composition, then exposure measures are an appropriate measure. To the extent we think that
segregation operates by exposing students to different school environments, however, evenness is
the appropriate measure, because under perfect evenness, all students experience the same
average school environments.
2.b. Trends in Black-White Segregation
2.b.1. Black-White Segregation in the Desegregation Era, 1954-1980
Black-white school desegregation trends can be divided roughly into two periods: the
period from 1954 through the 1970s, and the period from the 1980s to the present. In the first
period, black-white segregation declined dramatically, particularly in the South, though most of that
decline happened after 1968. Immediately following the Brown v. Board of Education decision in
1954, states and school districts did little to reduce racial segregation. In the South, many school
districts initially put into place so-called “freedom of choice” desegregation plans, which were
arguably designed largely to preserve racial segregation by putting the onus on black families to
enroll their children in white schools, an option unappealing to most black families given the
animosity of many white families to integration (Coleman et al. 1975; Welch & Light 1987;
Clotfelter 2004).
Not surprisingly, such plans achieved relatively little desegregation: Clotfelter (2004)
estimates that 81% of black students in the South and 72% of those in the border states still
attended majority black schools as of 1968; likewise, Orfield (2001) estimates that in the South
99% of blacks in 1964 and 86% in 1967 attended majority black schools. Segregation was nearly as
high in the rest of the country, by any measure. Nationally, 77% of black students attended
majority black schools in 1968 (Orfield 2001); over half of black students attended school where
90% or more of their classmates were black (Orfield 1983; Welch & Light 1987); and the average
black student was enrolled in a school where only 22% of students were white, despite the fact that
6
the public school student population was 79% white (Coleman et al. 1975). Studies using evenness
measures likewise report very high levels of segregation in 1968: the average within-district index
of dissimilarity between black and white public school students was about 0.80 (Logan & Oakley
2004; Johnson 2011); the average within-district variance ratio segregation index was 0.63
(Coleman et al. 1975). All of these measures exceed Massey & Denton’s (1989) threshold values for
“high segregation.”
In 1968, the Supreme Court’s Green decision (Green v. County School Board of New Kent
County, 391 US 430) required school districts to adopt more effective plans to achieve integration.
By the mid-1970s’s hundreds of school districts were subject to court-ordered desegregation plans
(Logan & Oakley 2004). As a result, school segregation levels declined substantially between 1968
and the mid-1970s. The average within-district variance ratio index dropped from 0.63 in 1968 to
0.37 in 1972; the black-white exposure index increased from 0.22 to 0.33 over the same time
period (Coleman et al. 1975), with the largest declines in segregation occurring in the South
(Coleman et al. 1975; Welch & Light 1987; Johnson 2011). The index of dissimilarity declined by
about 0.30 over the same time period, again declining more in the South than the North (Logan &
Oakley 2004; Welch & Light 1987). By 1980, only one-third of black students attended schools
where 90% or more of their classmates were black—still a substantial proportion but much lower
than in the late 1960s (Orfield 1983; Welch & Light 1987).
At the same time as within-district segregation was declining from 1968 to 1972, between-
district segregation was increasing (Coleman et al. 1975). This was particularly true in the North
where school districts are, on average, much smaller than districts in the South, where districts
often encompass whole counties. Coleman et al. (1975) find that within-district segregation
(defined by a measure of evenness) declined in every region from 1968 to 1972, particularly in the
South and Midwest, but that between-district segregation increased in every region. Particularly in
7
the non-South, declines in segregation within school districts were offset by increases between
districts.
2.b.2. Resegregation or Stalled Progress? Black-White Segregation Since 1980
The evidence is generally clear that school segregation between blacks and whites declined
substantially from 1968 to the mid-1970s and continued to modestly decline into the 1980s; this is
true whether one relies on measures of evenness or exposure. The evidence on trends in
segregation since the late 1980s, however, is less clear. On the one hand, Orfield and colleagues
have argued that the period from 1988 to the present is characterized by a gradual trend of
“resegregation” of black students (Orfield & Eaton 1997; Orfield 2001; Frankenburg & Lee 2002;
Frankenburg et al. 2003; Orfield & Lee 2007). To support this argument, they generally rely on
trends in exposure and isolation indices, reporting for example, that the black-white exposure index
was 0.27 in 2005, down substantially from its peak of 0.36 in 1988 and even lower than its level of
0.32 in 1970 (Orfield & Lee 2007; Orfield 2001; Frankenburg et al. 2003). Similarly, the proportion
of black students attending predominantly minority schools has risen from 63% in 1988 to 73% in
2005 (Orfield & Lee 2007).
In contrast, other scholars have argued that segregation has not risen significantly in the
last two decades. Using measures of evenness, Logan and colleagues find a very small increase in
black-white between-school segregation during the 1990s (Logan et al. 2002, 2008; Logan 2004;
Logan & Oakley 2004). Similarly, Stroub and Richards (2013) find that black-white segregation in
metropolitan areas rose very modestly from 1993-1998, but then declined from 1998-2009, for a
net decrease in average between-school metropolitan area segregation over the period from 1993-
2009. Black-white segregation between school districts also increased slightly during the 1990s
and remained higher than segregation within school districts (Clotfleter 1999; Reardon et al. 2000;
Logan et al. 2008; Logan & Oakley, 2004). During the 2000s, however, between-district racial
segregation declined slightly, but remains high (Stroub & Richards 2013).
8
Researchers have paid special attention to segregation trends in the South, given the
historically high levels of segregation and the focus of desegregation litigation on the region. Orfield
and colleagues argue that the resegregation of black students since 1988 is particularly pronounced
in the South and in the border states. By most measures, the South has been the least segregated
region of the country since the early 1970s, but it moved rapidly back to 1968 segregation levels (as
measured by black-white exposure) beginning in the late 1980s (Orfield and Lee 2007). Several
studies find that black-white segregation in the South increased during the 1990s, whether
measured using the exposure index or Theil’s entropy index, an evenness measure which assesses
segregation while taking demographic changes into account (Yun & Reardon 2002; Reardon & Yun
2003; Stroub & Richards 2013). The increase, however, is not large, and reversed following 1998
(Stroub & Richards 2013).
The debate about whether the last two decades can be characterized as a period of
“resegregation” largely hinges on whether one uses exposure or evenness measures of segregation.
The trends noted by Orfield and colleagues are due in part to changes in the racial composition of
the U.S. public school student population, which is substantially less white than it was 25 years ago.
Because of this, measures of black-white exposure would be expected to decline, even if the
reduction in white enrollments happened uniformly across all schools so that evenness measures
did not change (Logan 2004; Fiel 2013). Thus, it seems fair to say that the last 25 years have been
characterized by largely stable patterns of sorting of students among schools (unevenness) while
the racial/ethnic composition of the student population has changed substantially, a pair of trends
that yields declining black-white exposure measures but no significant change in evenness
measures of segregation. Whether this represents progress or stagnation depends on one’s theory
of how and why segregation matters.
2.c. Trends in Hispanic-White, Asian-White, and Multiracial Segregation
9
Given the historical context of the Brown case and its focus on black-white segregation, less
research has examined segregation among students of other races. Changing racial classifications,
particularly with regard to Hispanics, also limits the documentation of long-term trends in
segregation of other groups. As the student population has become more multiracial, new efforts
have been made to document segregation among all groups. Orfield and colleagues, again relying on
exposure measures, argue that Hispanic students have experienced continually increasing
segregation from whites since 1968, as Hispanic students’ exposure to white students has steadily
fallen since the late 1960s and representation in majority-minority schools has steadily risen
(Orfield 2001; Frankenburg & Lee 2002; Frankenburg et al. 2003; Orfield & Lee 2007). Evenness
measures of segregation, however, show only a very slight increase in Hispanic-white and Asian-
white segregation during the 1990s and 2000s (Logan et al. 2002; Stroub & Richards, 2013). The
discrepancy between these findings, again, is due to the difference in segregation measures used.
Three studies assess the trends in multiracial segregation in the last two decades (Reardon
et al. 2000; Stroub & Richards, 2013; Fiel 2013). Each uses an index (Theil’s H) that assesses the
evenness with which white, black, Hispanic, and Asian students are distributed among schools. The
studies conclude that segregation between whites and non-whites was flat or increased very
slightly during the 1990s while segregation among minority groups declined during this time.
However, from 1998-2009, segregation between whites and minorities declined modestly, while
segregation among minority groups continued to decline; as a result, multiracial segregation was
10% lower in 2009 than in 1993 (Stroub & Richards, 2013).
2.d. Trends in Economic Segregation
Many scholars have documented the high levels of poverty in majority-minority schools,
arguing that school segregation concentrates minority students in high-poverty schools, which tend
to have lower resources and student achievement (Orfield 2001; Frankenburg et al. 2003; Orfield &
Lee 2005, 2007; Logan et al. 2012; Saporito & Sohoni 2007). Orfield & Lee (2007) show that in
10
2005, the average black or Latino student attended a school in which 60% of students were poor;
the average white student attended a school in which a third of the students were poor.
While researchers note the link between racial and economic school composition, there is
surprisingly little research explicitly measuring economic segregation among schools. Studies of
residential income segregation show that neighborhood income segregation grew considerably
between 1970 and 2009 (Reardon and Bischoff 2011a; 2011b; Watson 2009; Jargowsky 1996).
Much of the growth in income segregation was due to the increasing segregation of the rich from all
other families. These trends would suggest that economic school segregation may have increased as
well over the last 40 years, since most children attend school relatively near their neighborhood.
Studies of school segregation, however, are limited by the fact that there is no systematic source of
detailed family income data at the school level. Instead studies of school segregation measure
income using free lunch eligibility, a very coarse measure of income that may obscure patterns of
segregation at the high or low ends of the income distribution. Nonetheless, studies using these
data show that economic segregation increased modestly in the 1990s, particularly in elementary
grades and in large school districts (Rusk 2001; Owens et al. 2013), but economic segregation did
not change appreciably in the 2000s (Owens et al. 2013). These patterns do not match the reported
neighborhood segregation trends (Reardon & Bischoff 2011a; 2011b), though it is not clear
whether that is due to the fact that they rely on a much coarser measure of income or because
school enrollment patterns have not mirrored neighborhood segregation patterns closely.
However, one other study (Altonji & Mansfield 2011) provides suggestive evidence that segregation
by family income between schools did indeed follow the neighborhood segregation trends: the
proportion of variance in family income between schools rose in the 1970s and 1980s but declined
in the 1990s.
Although it is difficult to measure trends in income segregation between schools, it is
possible to estimate levels of between-district segregation using Census data that tabulates the
11
number of school age children, by family income, enrolled in public school in each school district in
the U.S. Using these data, Owens and colleagues (Owens 2013; Owens et al. 2013) find that
between-district economic segregation among public school students increased during the 1990s
and the 2000s in three-quarters of the 100 largest metropolitan areas. This increase was largely
driven by increases in segregation in the top two-thirds of the income distribution. Taken as a
whole, the trends in income segregation suggest that students have grown more segregated
between districts, but segregation within school districts has remained relatively constant over the
last 20 years.
2e. Factors Shaping Trends in School Segregation
Court-ordered desegregation was the single most important factor shaping the rapid
declines in segregation in the 1960s and 1970s. Segregation declined sharply in school districts in
the years immediately following court orders and implementation of desegregation plans (Guryan
2004; Reber 2005; Johnson 2011; Lutz 2011). However, other factors mattered as well. Logan and
Oakley (2004) note that desegregation also occurred in many districts that did not have
desegregation plans in place. For example, in the South, the dissimilarity index fell from 0.72 to
0.30 in districts not covered by desegregation plans and from 0.87 to 0.47 in districts that were
subject to desegregation plans from 1968 to 2000. Therefore, declines in segregation during this
time also occurred in response to other federal government actions aimed at equal rights and racial
equality, districts preemptively undertaking voluntary desegregation plans before legislation
occurred, and district leaders finding desegregation to be a worthy social and educational goal (see
also Cascio et al. 2010). Even if districts were not subject to desegregation legislation, the shift in
the legal and social environment and enforcement by political leaders contributed to declining
segregation in nearly all districts.
Because court-ordered desegregation generally dealt solely with patterns of within-district,
between-school segregation, legal desegregation efforts were largely ineffective at reducing
12
between-district segregation. In 1974, the Supreme Court’s Milliken v. Bradley (418 U.S. 717)
decision ruled out court-ordered inter-district desegregation plans, unless it could be shown that
the state was responsible for between-district segregation patterns, a burden of proof difficult to
meet. This is one reason that today racial segregation is higher between districts than within
districts, between schools (Stroub & Richards 2013). Fiel (2013) shows that the high degree of
sorting by race between school districts is more consequential for minority student isolation and
exposure to whites than within-district sorting.
There is some evidence that desegregation efforts also contributed to increasing between-
district segregation, as a result of so-called “white flight”—the movement of white families to
districts with fewer blacks in order to avoid racially integrated schools (Coleman et al. 1975; Farley
et al. 1980; Rossell 1975; Wilson 1985). Although some of the decline in white enrollments in
desegregating districts can be attributed to declining white birth rates, several studies suggest that
white flight in response to desegregation also played a substantial role (Welch & Light 1987; Reber
2005). Reber (2013) shows that white enrollment losses reduced the effects of desegregation plans
by about one-third.
In addition to white flight to other districts, whites also left the public school system. In
response to desegregation in the 1960s and 1970s, white enrollment in private schools increased,
particularly in majority black school districts (Clotfelter 1976, 2004) Reardon and Yun (2003)
found that this pattern continued into the 1990s in the South; further, they find that the between-
district public school segregation was about 40% higher than residential segregation, as a result of
high rates of whites private school attendance in majority black districts. In contrast, Logan et al.
(2008), however, find mixed evidence that the availability of private schooling is associated with
racial segregation from 1970 to 2000.
Since the 1980s, several countervailing trends have operated to keep segregation levels
relatively stable. The changing legal context led to increases in segregation levels in some districts.
13
Between 1990 and 2010, hundreds of districts that had court-ordered desegregation plans were
released from court oversight (Reardon et al 2012). As a result, these districts became, on average,
increasingly segregated (U.S. Commission on Civil Rights, 2007; An & Gamoran 2009; Clotfelter et
al. 2006; Lutz 2011; Reardon et al. 2012). In addition, the Supreme Court’s 2007 decision in the
Parents Involved in Community Schools v. Seattle School District No. 1 (551 U.S. 701) outlawed the
use of students’ race in voluntarily-adopted school assignment plans, making it harder for districts
to voluntarily desegregation.
One potential countervailing force to this changing legal climate is the increased use of
socioeconomic-based student assignment plans (SBSAs), which attempt to create socioeconomic
integration in schools. Although there are some successful examples (Kahlenberg 2002, 2006),
most SBSAs have done little to reduce either socioeconomic or racial segregation (Flinspach et al.
2003; Reardon et al. 2006; Reardon & Rhodes 2011). The student assignment plans in place today,
then, are much weaker than desegregation plans of the 1960s and 1970s that substantially
integrated schools.
A more powerful countervailing force to the retreat from desegregation efforts is the
gradual decline in racial residential segregation. Black-white racial segregation has declined slowly
and steadily from 1980 to 2010; segregation between non-Hispanic whites and Hispanics and non-
Hispanic whites and Asians has remained fairly stable (and lower than black-white segregation)
during this time (Farley & Frey 1994; Logan et al. 2004; Logan & Stults 2011; Iceland & Sharp
2013). Because residential patterns partly determine school segregation patterns, this decline in
residential segregation has likely partially offset some of the increasing segregation due to the
decline in desegregation efforts. Nonetheless, although residential patterns are important, they are
not determinative of student body composition for several reasons. First, neighborhood and school
attendance zones map onto one another imperfectly. Second, many districts do not operate
neighborhood schools, instead offering assignment and choice plans through which students could
14
attend school outside their neighborhood. Third, some parents opt to send their child to private
school. Reardon & Yun (2003) provide evidence that residential and school segregation do not
necessarily follow one another: in the South, black-white neighborhood segregation declined in the
1990s while school segregation increased slightly in many Southern states and metro areas.
Finally, one reason that between-district segregation may have increased in recent decades
is that large-scale residential patterns, which particularly affect segregation between school
districts, has rose in the 1990s (Lee et al. 2008; Reardon et al. 2009). Consistent with this trend,
between-district racial segregation rose through the 1990s (Rivkin 1994; Clotfelter 2001; Reardon
et al. 2000; Stroub & Richards 2013).
3. Consequences of School Segregation
3.a. A Stylized Model of Segregation Effects on Students
Prior to reviewing the evidence on the consequences of segregation, it is useful to consider
the mechanisms through which school segregation may affect student outcomes. Here we lay out a
very general model for thinking about how segregation might affect students.
We can think of each school as having a set of resources that are beneficial to their enrolled
students. These resources may include the physical facilities of the school, the skills of the teachers
and staff, the school climate and curriculum, the social capital of the parents of the enrolled
students, and so on. To the extent that a student’s peers’ characteristics—such as their academic
skills, socioeconomic status, and race—affect his or her academic or social outcomes (including
attitudes, beliefs, friendship patterns, etc.), we can consider aggregate student characteristics as a
potential school resource as well. Suppose a student outcome 𝑌 is affected by the availability of
various school resources (denoted 𝑅1, … ,𝑅𝐾) and by other factors. Then we can write (assuming an
additive linear relationship between resources and outcomes):
15
𝑌 = �𝑎𝑘𝑅𝑘𝑠𝑘
+ 𝑒.
[1]
Here 𝑎𝑘 is the effect of school resource 𝑘 on student outcome 𝑌. The model is, of course,
oversimplified by its linear nature and assumption that resources have the same effects on all
students, but it is a useful stylized model for our purposes here.
Schools will, in general, differ in the degree to which they have access to various types of
resources, in part because some of the potential resources are tied to the student composition, and
in part because school districts and governments have some control over the allocation of certain
types of resources among schools (they may determine who teaches in which schools, or how
financial resources are distributed among schools). Moreover, the total amount of such resources
within a school system need not be fixed—states may allocate more or less money to schools;
districts may be more or less successful at recruiting skilled teachers; parents with resources and
social capital may move in or out of the district; and so on. In a general sense, then, segregation may
affect both the total quantity of a given resource within an educational system and the allocation of
the resource among schools. A stylized model of the association between the availability of resource
𝑘 in school 𝑠 could be written
𝑅𝑘𝑠 = 𝑏𝑘𝑉𝑑 + 𝑐𝑘𝑃𝑠 + 𝑢𝑘𝑠,
[2]
where 𝑃𝑠 is the proportion black (or proportion poor, or some other measure of school
composition) in school 𝑠 and 𝑉𝑑 is the variance ratio measure of segregation (a measure of
evenness; see Coleman, 1975) in the school district. We use the variance ratio for simplicity here, as
it makes the derivations below straightforward.
It is useful to consider, in concrete terms, what the coefficients in Equation [2] represent.
The coefficient 𝑏𝑘 indicates the relationship between the segregation level of the district and the
16
total quantity of resource 𝑅𝑘 available in the district. Because 𝑉𝑑 is constant across schools in a
district, the quantity 𝑏𝑘𝑉𝑑 is constant across schools for a given value of segregation. For example,
in the South, prior to the Brown and Green decisions, Southern states spent very little on black
schools relative to what they spent on white schools. Desegregation, however, led to rapid increases
in state spending on education, driven by white-controlled legislatures’ desire to ensure that white
students’ school quality did not decline with integration (Johnson 2011). In this case, the state
invested fewer total resources in the segregated school system than in the desegregated system,
implying that 𝑏𝑘 < 0 when 𝑅𝑘 measures financial resources. However, segregation might also lead
to higher total available resources. For example, if segregation between schools causes more high-
income families to remain in a school district, and if we think of such families as a resource to the
schools their children attend (perhaps because they have more political power, on average, or
because they serve as role models for their children and their children’s classmates, or because they
are more likely to have time to volunteer or be otherwise involved in the school), then segregation
may lead to greater total resources in the district. In this case, 𝑏𝑘 > 0 when 𝑅𝑘 measures parental
social and economic capital.
A second way that segregation may affect students is by affecting how the district’s
available resources are distributed among students. This is described by the coefficient 𝑐𝑘 in
Equation [2], the association between school racial composition (proportion black) and the
availability of resource 𝑅𝑘 in a district. For example, suppose that, within a district, more skilled
teachers are more likely to teach in low-poverty schools than in high-poverty schools (perhaps
because higher-income parents are able to persuade district leaders to assign certain teachers to
their children’s schools, or because high-poverty schools are less able to attract and retain the most
skilled teachers). If this is true, then segregation may heighten the disparity in the average quality
of teachers available to poor and non-poor students within a district, implying 𝑐𝑘 < 0 if 𝑅𝑘
measures teacher quality. Similarly, if peers affect one another’s academic or social outcomes, then
17
segregation may lead to a more unequal distribution of peer resources among schools: poor
students will have less exposure to higher-achieving classmates (given the correlation between
income and academic skills prior to school entry) than will non-poor students, again implying
𝑐𝑘 < 0 when 𝑅𝑘 measures average student academic skills. Conversely, if districts react to
socioeconomic segregation among schools by allocating more of their resources to high-poverty
schools, then segregation may (in principle) lead to a positive disparity (more resources in the
school of the average poor student than the average non-poor student). In this case, 𝑐𝑘 > 0.
Note that in the above discussion, our point is not to make claims regarding whether and
how specific resources affect student outcomes, nor to assess how the quantity or allocation of
resources is affected by segregation. Rather, our point here is to suggest two general classes of
mechanisms through which segregation may affect student outcomes: by affecting the total pool of
available resources in a school district (in which case 𝑏𝑘 ≠ 0); and/or by affecting the distribution
of available resources among schools (in which case 𝑐𝑘 ≠ 0).
From the model above, we can derive several useful relationships. First, note that Equations
[1] and [2] imply that the average outcome in the district will be
𝐸[𝑌|𝑑] = �𝑎𝑘𝐸[𝑅𝑠𝑘|𝑑]𝑘
+ 𝐸[𝑒|𝑑]
= �𝑎𝑘(𝑏𝑘𝑉𝑑 + 𝑐𝑘𝐸[𝑃𝑠|𝑑])𝑘
= �𝑎𝑘𝑏𝑘𝑉𝑑 + 𝑎𝑘𝑐𝑘𝑃𝑑𝑘
= 𝑉𝑑�𝑎𝑘𝑏𝑘𝑘
+ 𝑃𝑑�𝑎𝑘𝑐𝑘𝑘
= 𝑏∗𝑉𝑑 + 𝑐∗𝑃𝑑 ,
[3]
where 𝑃𝑑 is the proportion black in the district as a whole and where 𝑏∗ = ∑ 𝑎𝑘𝑏𝑘𝑘 and
𝑐∗ = ∑ 𝑎𝑘𝑐𝑘𝑘 . Note that in this stylized model, the average outcome 𝑌 in a district will be a function
18
of its segregation level. For simplicity here, assume that the racial composition of a district is held
constant while its segregation level is altered; then 𝑏∗ is the total effect of segregation on student
outcomes. This total effect is the sum of the effects of each resource 𝑘 that is both affected by
segregation (i.e., 𝑏𝑘 ≠ 0) and that affects student outcome 𝑌 (i.e., , 𝑎𝑘 ≠ 0). If segregation increases
the availability of some resources and decreases the availability of others, then some pathways
through which segregation affects outcomes may partially cancel each other out; that is, 𝑏∗ = 0
does not imply that segregation has no effect on resources or that resources do not affect
achievement. The key insight provided by this model is that we can think of the total effect of
segregation as the sum of a set of mechanisms. Understanding if and how segregation affects
student outcomes depends in part on knowing how segregation affects school district resources
and how school resources affect students.
Equation [3] describes the relationship between segregation and average student outcomes.
Next we consider how segregation affects disparities in school resources and student outcomes.
Equation [2] implies that the difference in school resources in the schools of black and white
students will be
𝐸[𝑅𝑘𝑠|𝑏𝑙𝑎𝑐𝑘] − 𝐸[𝑅𝑘𝑠|𝑤ℎ𝑖𝑡𝑒] = (𝑏𝑘𝑉𝑑 + 𝑐𝑘𝐸[𝑃𝑠|𝑏𝑙𝑎𝑐𝑘]) − (𝑏𝑘𝑉𝑑 + 𝑐𝑘𝐸[𝑃𝑠|𝑤ℎ𝑖𝑡𝑒])
= 𝑐𝑘�𝑃�𝑠𝑏𝑙𝑎𝑐𝑘 − 𝑃�𝑠𝑤ℎ𝑖𝑡𝑒�
= 𝑐𝑘𝑉𝑑,
[4]
where 𝑃�𝑠𝑏𝑙𝑎𝑐𝑘 and 𝑃�𝑠𝑤ℎ𝑖𝑡𝑒 are the average proportion black in the schools of black and white
students, respectively (these are exposure indices). Conveniently, the difference 𝑃�𝑠𝑏𝑙𝑎𝑐𝑘 − 𝑃�𝑠𝑤ℎ𝑖𝑡𝑒 is
equal to the variance ratio index measure of segregation 𝑉𝑑. Therefore, the difference in the
exposure of black and white students to school resource 𝑅𝑘 is determined by the segregation level
of the district and the extent to which school racial composition affects the allocation of 𝑅𝑘 among
schools (𝑐𝑘).
19
Finally, note that Equation [1] implies that the black-white difference in average student
outcome 𝑌 will be
𝐸[𝑌|𝑏𝑙𝑎𝑐𝑘] − 𝐸[𝑌|𝑤ℎ𝑖𝑡𝑒] = �𝑎𝑘(𝐸[𝑅𝑘𝑠|𝑏𝑙𝑎𝑐𝑘] − 𝐸[𝑅𝑘𝑠|𝑤ℎ𝑖𝑡𝑒])𝑘
= �𝑎𝑘(𝑐𝑘𝑉𝑑)𝑘
= 𝑉𝑑�𝑎𝑘𝑐𝑘𝑘
= 𝑐∗𝑉𝑑 .
[5]
Equation [5] makes clear that segregation will affect racial disparities in student outcomes if
𝑐∗ = ∑ 𝑎𝑘𝑐𝑘𝑘 ≠ 0. That is, if school racial composition affects the allocation of resources among
schools, and if those resources affect students, then segregation will lead to disparities in student
outcomes.
This stylized model formalizes the two mechanisms through which segregation may affect
student outcomes that we described above. First, if segregation changes the total pool of resources
available to a school district, it will affect average student outcomes (so long as those resources
affect student outcomes). And second, if school resources are allocated among schools in ways
correlated with school racial composition, then segregation will lead to racial disparities in the
outcome 𝑌 (again, so long as those resources affect student outcomes). Of course, this stylized
model is overly simple—it assumes homogeneous, linear, additive effects of segregation and racial
composition on school resources and of school resources on student outcomes—but it is
nonetheless useful for clarifying the parameters of interest in understanding the effects of
segregation.
3.b. Evidence on the Consequences of School Segregation
20
As is evident in the stylized model of segregation effects above, there are a number of
parameters relevant to understanding the effects of segregation. The total effects of segregation on
average outcomes and on disparities in outcomes are captured by the parameters 𝑏∗ and 𝑐∗. It is
useful to estimate these parameters, because they describe the total effects of segregation on
average outcomes and outcome disparities, respectively. The individual 𝑎𝑘, 𝑏𝑘, and 𝑐𝑘 parameters
are also of interest, of course, because they describe the specific pathways through which
segregation affects outcomes; knowing these parameters is useful from both the perspective of
sociological theory and social policy. Direct estimation of any of these parameters, however, is
complicated by the fact that school resources, segregation levels, and school racial composition
levels are rarely ignorably assigned. There are, however, a small number of studies that provide
credible estimates of some of these parameters.
Several studies estimate the impacts of school segregation by examining how black and
white students’ outcomes changed during the era of school desegregation. These studies use the
exogenous variation in timing of desegregation court orders or implementation to estimate the
effects of desegregation on students’ outcomes and disparities in those outcomes (i.e., they estimate
𝑏∗ and/or 𝑐∗). Guryan (2004) finds that desegregation led to a decline in black dropout rates during
the 1970s of 2-3 points, accounting for about half the decline in the black dropout rate during this
time. Johnson (2011) finds that blacks’ odds of graduating from high school increased by about 1
percentage point and their educational attainment increased by about 1/10 of a year for every
additional year they were exposed to a school desegregation order. Neither study finds significant
effects on the educational attainment of whites, suggesting that school desegregation was not
harmful for whites. In other words, they suggest that desegregation had a positive effect on average
attainment and reduced racial attainment disparities. Other studies also find a positive relationship
between school integration and educational outcomes for blacks (Boozer et al. 1992; Reber 2010).
21
In addition to educational attainment, scholars have examined the impacts of desegregation
on later life outcomes (see Wells and Crain, 1994, for a review). Several studies show that increased
exposure to school desegregation improved black adult males’ earnings, reduced the odds of
poverty, and increased the odds of working white-color jobs (Crain and Strauss 1985; Boozer et al.
1992; Ashenfelter et al. 2006; Johnson 2011).
Other studies find effects of desegregation on social outcomes like criminality and health.
Exposure to desegregation orders reduces the probability of men’s deviant behavior, homicide
victimization, arrests, and incarceration (Johnson 2011; Weiner, Lutz, and Ludwig 2009) and
improves adult health (Johnson 2011) Taking a multi-generational view, Johnson (2013) found
that school desegregation affects not only those exposed to it, but also their children and
grandchildren. Exposure to school desegregation positively affects the reading and math test
scores, educational attainment, college quality, and racial diversity at college of the “children and
grandchildren of Brown,” with parent and grandparent educational attainment serving as a key
mechanism.
Another way to assess the impact of school desegregation on student outcomes is to
examine what happens once court orders have been dismissed. Lutz (2011) found that the
dismissal of court-ordered desegregation plans increased black dropout rates outside the South,
and Saatcioglu (2010) found that the end of desegregation policy in Cleveland led to higher dropout
rates among black and Hispanic students. Vigdor (2011), however, found that the black-white test
score gap did not widen among elementary schools following the end of busing in Charlotte-
Mecklenburg. It could be the case that desegregation affects test scores and dropout differently, as
little research examined test scores using variation in desegregation orders due to data limitations.
Finally, a few studies have examined the relationship between city or metro area
segregation levels and test score gaps. Card and Rothstein (2007) examine the effects of
neighborhood and school segregation on the black-white test score gap and find that the black-
22
white test score gap is higher in more segregated cities but that school segregation has no
independent effect when neighborhood segregation is accounted for. Mayer (2002) finds that
neighborhood economic segregation, which may be correlated with school economic segregation,
increases educational attainment for high-income students but slightly reduces low-income
children’s attainment, with little net effect overall.
The studies reviewed here often try to test mechanisms that explain why desegregation
improved black students’ outcomes. Most focus on how segregation shapes the distribution of
resources rather than the overall level available in the district (that is, they test whether 𝑐𝑘 = 0).
Generally, they find that desegregation in the South equalized the length of school year, student-
teacher ratios, teacher quality, and per-pupil expenditures that the average black and white student
experienced (Ashenfelter et al. 2006; Card & Krueger 1992; Guryan 2004; Johnson 2011; Reber
2010). Several studies also consider peer effects, arguing that exposure to white peers may benefit
blacks because whites tended to be higher-SES, and they might be more highly motivated or higher
achieving (Guryan 2004; Reber 2010; Ashenfelter et al. 2006; Saatcioglu 2010).
Finally, researchers acknowledge that the act of desegregation itself may have helped black
students feel more enfranchised, optimistic about their futures, and dedicated to their studies,
perhaps also increasing parental involvement, all of which could improve their educational
outcomes (Ashenfelter et al. 2006; Guryan 2004). Desegregation may also increase the expectations
of parents, teachers, and other adults who interact with black children (Johnson 2011).
3.c. Evidence on the Consequences of School Composition
A number of studies have tried to estimate the effect of school racial composition on student
outcomes, as a way of understanding the effects of segregation (see Hallinan 1998, Vigdor and
Ludwig 2008, Mickelson and Bottia 2009 for reviews). This can be problematic, however, because
racial composition may not directly affect student outcomes, but may operate through its effect on
23
other resources. To see this, consider the result of substituting equation [2] into Equation [1]
above:
𝑌 = �𝑎𝑘(𝑏𝑘𝑉𝑑 + 𝑐𝑘𝑃𝑠 + 𝑢𝑘𝑠)𝑘
+ 𝑒
= 𝑏∗𝑉𝑑 + 𝑐∗𝑃𝑠 + 𝑒∗,
[6]
where 𝑒∗ = ∑ 𝑎𝑘𝑢𝑘𝑠𝑘 + 𝑒. Regressing 𝑌 on school racial composition (𝑃𝑠), holding segregation
constant, will yield an estimate of 𝑐∗, the total effect of racial composition on achievement, which is
identical to the effect of segregation on racial disparities in outcomes. However, because schools are
rarely assigned to have different racial compositions, the estimation of 𝑐∗ from [8] will generally
lead to biased estimates, unless the regression model includes adequate control variables or a
quasi-experimental design is used to identify 𝑐∗.
Studies that include control variables in Equation [8], however, run the risk of increasing
the bias in the estimates of 𝑐∗, however, if the covariates are affected by racial composition. To see
this, consider the regression model below, where 𝑗 indexes various school covariates (the 𝑋𝑗’s),
some of which may be resources that affect student outcomes included in Equation [1]:
𝑌 = 𝑏′𝑉𝑑 + 𝑐′𝑃𝑠 + �𝑎𝑗′𝑋𝑗𝑠𝑗
+ 𝑒′.
[7]
Allowing each 𝑋𝑗 to be a function of district segregation, school racial composition, and some other
factors uncorrelated with school composition, we can rewrite [7] in the same form as Equation [6]:
𝑌 = 𝑏′𝑉𝑑 + 𝑐′𝑃𝑠 + �𝑎𝑗′�𝑏𝑗𝑉𝑑 + 𝑐𝑗𝑃𝑠 + 𝑢𝑗𝑠�𝑗
+ 𝑒′
= �𝑏′ + �𝑎𝑗′𝑏𝑗𝑗
�𝑉𝑑 + �𝑐′ + �𝑎𝑗′𝑐𝑗𝑗
�𝑃𝑠 + 𝑒′∗
= 𝑏∗𝑉𝑑 + 𝑐∗𝑃𝑠 + 𝑒∗
24
[8]
Equation [8] shows that the coefficient on racial composition in Equation [7] will be equal to
𝑐′ = 𝑐∗ −�𝑎𝑗′𝑐𝑗𝑗
.
[9]
Thus, fitting Equation [7] will not yield an unbiased estimate of 𝑐∗ unless none of the school
covariates included in [7] are associated with school racial composition (i.e., 𝑐𝑗 = 0 for all 𝑋𝑗 in the
model). Put differently, controlling for downstream mediators of school composition will lead to
biased estimates of the effects of school composition. Because it is not always clear which variables
should be considered correlates of composition (which should be controlled for) and which should
be considered downstream mediators of the effects of school composition (which should not be
controlled for), there is an inherent ambiguity in regression-based estimates of the effects of school
composition. In most cases, neither estimates of 𝑐∗ from Model [6] or of 𝑐′ from Model [7] can be
considered to have a strong causal warrant, and should not be used to infer the effects of
segregation.
As a result of these challenges, there are relatively few studies that provide compelling
estimates of the effects of school composition. Several studies, however, use research designs that
provide some plausible exogeneity in the sorting of students into schools. First, two studies use
data from the Texas School Project and take advantage of plausibly random variation in cohort
demographics over time. They find that a reduction in black enrollment increases reading and math
test scores for black students and does not harm whites’ test scores (Hanushek et al.2002; Hoxby
2000). Second, several studies take advantage of random assignment of children to schools or
neighborhoods to examine how changing school composition may affect educational outcomes.
Sanbonmatsu et al. (2006) find no significant effects on test scores among children whose families
received housing vouchers to be used in low poverty neighborhoods. Few children whose families
25
received housing vouchers and moved changed schools, however, so the study cannot test the
impacts of exogenous changes in school environments on educational achievement. Schwartz
(2010) takes advantage of the fact that Montgomery County randomly assigns students in public
housing to different schools and compares the performance of those who attended the district’s
most versus least advantaged schools. She finds that by their fifth year of elementary school
students from public housing in low-poverty elementary schools had significantly higher scores in
math and reading than equally poor students assigned to high-poverty schools. These positive
impacts accumulate over time—by the seventh year of school, low-income students in low-poverty
schools outperformed their peers at high-poverty schools by 0.4 standard deviations in math and
0.2 standard deviations in reading. This study provides the best experimental evidence that school
economic composition affects test scores.
4. Conclusion
Although the 1954 Brown decision is rightly hailed as the most significant Supreme Court
decision concerning schools in U.S. history, it had little immediate impact on school segregation.
Indeed, the most significant changes in school segregation in the United Stated did not begin until
1968, following the Green decision, after which black-white school segregation declined sharply
over a period of 5-10 years. Over the last 25 years, however, and despite claims of re-segregation
on the one hand (Orfield 2001; Orfield & Lee 2007) and “the end of the segregated century” on the
other (Vigdor and Glaeser 2012), school racial segregation has changed very little. There have been
modest decreases in the exposure of minorities to whites, but these have been driven primarily by
demographic changes in the school-age population (Fiel 2013; Logan 2004). Segregation measured
as unevenness has declined very modestly over the last two decades.
One of the conclusions evident from a review of the research on trends in segregation is that
we know a great deal about trends in racial segregation among K-12 public schools, but relatively
26
little about a number of other dimensions of segregation. First, due to data limitations, we know
relatively little about trends in economic segregation in the last two decades, and virtually nothing
about economic segregation prior to 1990. Second, very few studies consider trends in segregation
in postsecondary education (two recent exceptions are Hinrichs 2012; Carnevale & Strohl 2013) or
in pre-school settings. Third, few studies examine trends in segregation between private and public
schools or among private schools (but see Reardon & Yun 2002; Fiel 2013). And fourth, we have
relatively little research on patterns and trends of within-school segregation, though studies of
tracking (e.g., Oakes 1985; Lucas 1999), teacher assignment (Clotfelter et al. 2006: Kalogrides &
Loeb 2013) and students’ friendship networks (Tatum 1997; Moody 2001; Flashman 2013; Grewal
2012; Fletcher et al. 2013) suggest high levels of within-school segregation. We know very little
about how these patterns have changed over time (though see Conger 2005, for within-school
segregation trends in New York City). In each of these areas, research to identify the key patterns
and trends would be very useful for understanding the extent to which schools have become more
or less segregated along many dimensions.
Research on patterns and trends in segregation are generally motivated by a concern that
segregation leads to racial and socioeconomic disparities in educational outcomes. Surprisingly,
however, the sociological literature appears to lack a detailed and comprehensive theoretical model
(or models) of exactly how segregation might affect educational and social outcomes. As a result,
many studies estimate different parameters, all under the rubric of understanding the effects of
segregation. Given the theoretical confusion in the literature, one of our aims in this review was to
try to articulate a very general and stylized model for understanding how segregation might affect
student outcomes and to characterize the types of parameters of interest in the issue. While our
model is certainly incomplete and over-simplified, it may provide a useful framework for future
theoretical specification.
27
Our model suggests that two types of parameters are of particular interest in the study of
segregation: estimates of the effect of segregation per se on educational outcomes (what economists
call “reduced form” estimates, and what sociologists think of as “total effects”); and estimates of the
parameters defining the mechanisms through which segregation operates. To date, the research
literature has been more successful at providing the first type of estimates, particularly in
relationship to the effects of the segregation/desegregation in the 1960s and 1970s. Studies of this
type show that desegregation led to improvements in the educational outcomes of black students
while not harming those of white students. Studies of more recent segregation provide, however,
weak and mixed evidence on the degree to which segregation is linked to achievement gaps today.
Part of the reason for this disparity may be that some of the component mechanisms connecting
segregation to educational outcomes have changed. Johnson (2011) and others argue that pre-
1968 segregation was linked to substantial black-white inequality in school resources (inequalities
that were substantially reduced by desegregation). Segregation today is not strongly linked to
school resource inequality (in terms of financial resources). If segregation in the pre-Green era
operated primarily through its effects on the inequality of school funding, it may be less
consequential in the modern era of smaller funding disparities.
This last point indicates the need for much more theoretical and empirical understanding of
the mechanisms through which segregation affects student outcomes. To this end, our conceptual
model suggests that future research should focus on three types of questions to clarify the
mechanisms through which segregation operates. First, how does the segregation of a schooling
system affect the total quantity of available resources in the system? The list of resources of
interest here should include not only financial resources, but a wide range of other resources,
including human capital, social capital, peer characteristics, access to social networks,
neighborhood conditions, and so on. Second, how are resources distributed among schools in
relation to schools’ racial and socioeconomic composition? And third, how do these school
28
resources affect students’ educational outcomes? These are not simple questions to answer, of
course. Nonetheless, identifying and understanding the mechanisms through which segregation
affects (or doesn’t affect) students will likely do much more than will additional measurement of
trends and patterns to advance our understanding of why and how segregation matters.
29
Literature Cited
Altonji JG, Mansfield RK. 2011. The role of family, school, and community characteristics in
inequality in education and labor-market outcomes. In Whither Opportunity: Rising Inequality,
Schools, and Children’s Life Chances, ed. GJ Duncan, RJ Murnane, pp. 339-58. New York: Russell Sage
Found.
An BP, Gamoran A. 2009. Trends in school racial composition in the era of unitary status. In From
the Courtroom to the Classroom: The Shifting Landscape of School Desegregation, ed. CE Smrekar, EB
Goldring, pp. 19-48. Camb., MA: Harv. Educ. Press.
Ashenfelter O, Collins WJ, Yoon, A. 2006. Evaluating the role of Brown v. Board of Education in
school equalization, desegregation, and the income of African Americans. Am. Law and Econ. Rev.
8(2):213-48.
Bischoff K. 2008. School district fragmentation and racial residential segregation: how do
boundaries matter? Urban Aff. Rev. 44(2):182-217.
Boozer MA, Krueger AB, Wolkon S. 1992. Race and school quality since Brown vs. Board of
Education. Work. Pap. No. w4109. Natl. Bur. of Econ. Res.
Brown v. Board of Education, 347 U.S. 483, 74 S. Ct. 686, 98 L. Ed. 873 (1954).
Card D, Krueger AB. 1992. School quality and black-white relative earnings: a direct assessment.
Quar. J. Econ. 107(1):151-200.
30
Card D, Rothstein J. 2007. Racial segregation and the black-white test score gap. J. Publ Econ. 91(11-
12):2158-84.
Carnevale AP, Strohl J. 2013. White flight goes to college. Poverty & Race Res. Action Council 22(5):
1-2, 12-13.
Cascio E, Gordon N, Lewis E, Reber S. 2010. Paying for progress: conditional grants and the
desegregation of southern schools. Quar. J. Econ. 125(1): 445-82.
Charles, CA. 2003. The dynamics of racial residential segregation. Annu. Rev. Sociol. 29:167-207.
Clotfelter CT. 1976. School desegregation, ‘tipping,’ and private school enrollment. J. of Hum. Resour.
22:29-50.
Clotfelter, CT. 1999. Public school segregation in metropolitan areas. Land Econ. 75(4):487-504.
Clotfelter, CT. 2001. Are whites still fleeing? Racial patterns and enrollment shifts in urban public
schools, 1987-1996. J. of Policy Anal. and Manag. 20(2):199-221.
Clotfelter, CT. 2004. After Brown: The Rise and Retreat of School Desegregation. Princet., NJ: Princet.
Univ. Press.
Clotfelter CT, Ladd HF, Vigdor JL. 2006. Federal oversight, local control and the specter of
resegregation in southern schools. Am. Law and Econ. Rev. 8:1-43.
31
Clotfelter CT, Ladd HF, Vigdor JF. 2006. Teacher-student matching and the assessment of teacher
effectiveness. J. Hum. Resour. 4:778-820.
Coleman JS, Kelly SD, Moore JA. 1975. Trends in School Segregation, 1968-73. Wash., DC: Urban Inst.
Conger D. 2005. Within-school segregation in an urban school district. Educ. Eval. And Policy Anal.
27(3):225-44.
Crain RL, Strauss J. 1985. School desegregation and black occupational attainment: results from a
long-term experiment. Center for Soc. Org. of Schools. Johns Hopkins Univ.
Duncan OD, Duncan B. 1955. A methodological analysis of segregation indexes. Am. Sociol. Rev.
20:210-17.
Farley R, Frey WH. 1994. Changes in the segregation of whites from blacks during the 1980s: small
steps towards a more integrated society. Am. Sociol. Rev. 59:23-45.
Farley R, Richards T, Wurdock C. 1980. School desegregation and white flight: an investigation of
competing models and their discrepant findings. Sociol. of Educ. 53(3):123-39.
Fiel, JE. 2013. Decomposing school resegregation: social closure, racial imbalance, and racial
isolation. Am. Sociol. Rev.78(5):828-48.
Flashman J. 2013. That’s what friends are for? The role of friends in the maintenance of
32
socioeconomic achievement gaps. Unpublished Manuscr.
Fletcher J, Ross SL, Zhang Y. 2013. The determinants and consequences of friendship composition.
Work.Pap. No. 19215. Natl. Bur. of Econ. Res.
Flinspach SL, Banks KE, Khanna R. 2003. Socioeconomic integration policy as a tool for diversifying
schools: promise and practice in two large school systems. Presented at the Color Lines Conf., Harv.
Univ., Camb., MA.
Frankenberg E, Lee CM, Orfield G. 2003. A multiracial society with segregated schools: are we losing
the dream? Rep. The Civ. Rights Proj., Harv. Univ., Camb., MA.
Frankenburg E, Lee CM. 2002. Race in American public schools: rapidly resegregating school
districts. Rep. The Civ. Rights Proj., Harv. Univ., Camb., MA.
Glaeser E, Vigdor J. 2012. The end of the segregated century: racial separation in America’s
neighborhoods, 1890-2010. Civic Rep. No. 66. Manhattan Inst. for Pol. Res.
Green v. County School Board of New Kent County, 391 U.S. 430, 88 S. Ct. 1689, 20 L. Ed. 2d 716
(1968).
Grewal, ET. 2012. Exploring socioeconomic friendship segregation in schools. Unpublished
Manuscr.
Guryan J. 2004. Desegregation and black dropout rates. Am. Econ. Rev. 94(4):919-43.
33
Hallinan, MT. 1998. Diversity effects on student outcomes: social science evidence. Ohio State Law J.
59:733-54.
Hanushek, E, Kain, JF, Rickin S. 2009. New evidence about Brown v. Board of Education: the complex
effects of school racial composition on achievement. J. Lab. Econ. 27(3): 349-83.
Hinrichs P. 2012. An empirical analysis of racial segregation in higher education. Unpublished
manuscr.
Hoxby, C. 2000. Peer effects in the classroom: learning from gender and race variation. Work. Pap.
No. 7867. Natl. Bur. of Econ. Res.
Iceland J, Sharp G. 2013. White residential segregation in U.S. metropolitan areas: conceptual issues,
patterns, and trends from the U.S. census, 1980 to 2010. Popul. Res. and Policy Rev.32:663-86.
James DR, Taueber K. 1985. Measures of segregation. Sociol. Methodology 1-32.
Jargowsky PA.1996. Take the money and run: economic segregation in U.S. metropolitan areas. Am.
Sociol. Rev. 61:984-98.
Johnson RC. 2011. Long-run impacts of school desegregation & school quality on adult attainments.
Work. Pap. No. 16664. Natl. Bur. of Econ. Res.
Johnson RC. 2013. The grandchildren of brown: the long legacy of school desegregation.
34
Unpublished manuscr.
Kahlenberg RD. 2002. Economic school integration: an update. The Century Found. Issue Brief
Series, The Century Found., New York, NY.
Kahlenberg RD. 2006. Socioeconomic school integration. North Carolina Law Rev. 85: 1545-94.
Kalogrides D, Loeb S. 2013. Different teachers, different peers: the magnitude of student sorting
within schools. Educ. Research. 42(6): 304-16.
Lee B, Reardon SF, Firebaugh G, Farrell C, Matthews SA, O’Sullivan D. 2008. Beyond the census tract:
patterns and determinants of racial segregation at multiple geographic scales. Am. Sociol. Rev.
73(5):766-91.
Lieberson, S, Carter, DK. 1982. A model for inferring the voluntary and involuntary causes of
residential segregation. Demography 19(4): 511-26.
Logan J. 2004. Resegregation in American public schools? Not in the 1990s. Rep. Lewis Mumford
Cent. for Comparative Urban and Regional Res., SUNY Albany.
Logan JR, Oakley, J. 2004. The continuing legacy of the Brown decision: court action and school
segregation, 1960-2000. Rep. Lewis Mumford Cent. for Comparative Urban and Regional Res., SUNY
Albany.
35
Logan JR, Stults, BJ. 2011. The persistence of segregation in the metropolis: new findings from the
2010 census. Census Brief. Proj. US2010.
Logan JR, Minca E, Adar S. 2012. The geography of inequality: why separate means unequal in
American public schools. Sociol. of Educ. 85:287-301.
Logan JR, Oakley D, Stowell J. 2002. Choosing segregation: racial imbalance in American public
schools, 1990-2000. Rep. Lewis Mumford Cent. for Comparative Urban and Regional Res., SUNY
Albany.
Logan JR, Oakley D, Stowell J. 2008. School segregation in metropolitan regions, 1970-2000: the
impacts of policy choices on public education. Am. J. of Sociol. 113(6):1611-44.
Logan JR, Stults BJ, Farley R. 2004. Segregation of minorities in the metropolis: two decades of
change. Demography 41(1):1-22.
Lucas SR. 1999. Tracking Inequality: Stratification and Mobility in American High Schools. New York:
Teachers Coll. Press.
Lutz B. 2011. The end of court-ordered desegregation. Am. Econ. J.: Econ. Policy 3(2):130-68.
Massey DS, Denton NA. 1988. The dimensions of residential segregation. Soc. Forces 67:281-315.
Massey DS, Denton NA. 1989. Hypersegregation in U.S. metropolitan areas: black and Hispanic
segregation along five dimensions. Demography 26:373-93.
36
Mayer, SE. 2002. How economic segregation affects children’s educational attainment. Soc. Forces
81(1): 153-76.
Mickelson RA. 2001. Subverting Swann: first- and second-generation segregation in the Charlotte-
Mecklenburg schools. Am. Educ. Res. J. 38:215-52.
Mickelson, RA, Bottia, M. 2009. Integrated education and mathematics outcomes: a synthesis of
social science research. North Carol. Law Rev. 88:993-1090.
Milliken v. Bradley, 418 U.S. 717, 94 S. Ct 3112, 41 L. Ed. 2d 1069 (1974)
Moody J. 2001. Race, school integration, and friendship segregation in America. Am. J. of Sociol.
107(3):679-716.
Oakes J. 1985. Keeping Track: How Schools Structure Inequality. New Haven, CT: Yale Univ. Press.
Orfield G. 1983. Public School Desegregation in the United States, 1968-1980. Wash., DC: Joint Cent.
for Polit. Studies.
Orfield, G. 1995. Public opinion and school desegregation. Teach. Coll. Rec. 96(4): 654-70.
Orfield G. 2001. Schools more separate: consequences of a decade of resegregation. Rep. The Civ.
Rights Proj., Harv. Univ.
37
Orfield G, Lee CM. 2005. Why segregation matters: poverty and educational inequality. Rep. The Civ.
Rights Proj., Harv. Univ.
Orfield G, Lee CM. 2007. Historic reversals, accelerating resegregation, and the need for new
integration strategies. Rep. The Civ. Rights Proj./Proyecto Derechos Civiles, UCLA.
Orfield G, Eaton SE. 1997. Dismantling Desegregation: The Quiet Reversal of Brown v. Board of
Education. New York: The New Press.
Owens A. 2013. Economic segregation between school districts, 1990 to 2010. Unpublished
manuscr.
Owens A, Reardon SF, Jencks C. 2013. Trends in family income segregation among schools, 1970 to
2010. Presented at Income, Inequal., and Educ. Success Conf., Stanford Univ.
Parents Involved in Community Schools v. Seattle School District No. 1. 551 U.S. 701, 127 S. Ct. 2738;
75 U.S.L.W. 4577; 20 Fla. L. Weekly Fed. S 490 (2007)
Reardon SF, Bischoff K. 2011a. Income inequality and income segregation. Am. J. of Sociol.
116(4):1092-1153.
Reardon SF, Bischoff K. 2011b. Growth in the residential segregation of families by income, 1970-
2009. US2010 Proj. Report. Brown Univ. and Russell Sage Found.
Reardon SF, Rhodes L. 2011. The effects of socioeconomic school integration plans on racial school
38
desegregation. In Integrating Schools in a Changing Society: New Policies and Legal Options for a
Multiracial Generation, ed. E Frankenburg, E DeBray, pp. 187-208. Chapel Hill: Univ. of North Carol.
Press.
Reardon SF, Yun JT. 2002. Segregation in church and state? Vouchers and racial enrollment patterns
among public and private schools. Teach. Coll. Rec. Online Edition
http://www.tcrecord.org/Content.asp?ContentID=10968
Reardon SF, Yun JT. 2003. Integrating neighborhoods, segregating schools: the retreat from school
desegregation in the south, 1990-2000. North Carolina Law Rev. 81(4):1563-96.
Reardon SF, Yun JT, Eitle TM. 2000. The changing structure of school segregation: measurement
and evidence of multiracial metropolitan-area school segregation, 1989-1995. Demography
37(3):351-64.
Reardon SF, Yun JT, Kurlaender M. 2006. Implications of income-based school assignment policies
for racial school segregation. Educ. Eval. and Policy Anal. 28(1):49-75.
Reardon SF, Grewal ET, Kalogrides D, Greenberg E. 2012. Brown fades: the end of court-ordered
school desegregation and the resegregation of American public schools. J. of Policy Anal. and Manag.
31(4):876-904.
Reardon SF, Farrell CR, Matthews SA, O’Sullivan D, Bischoff K, Firebaugh G. 2009. Race and space in
the 1990s: changes in the geographic scale of racial residential segregation, 1990-2000. Soc. Sci. Res.
38:55-70.
39
Reber SJ. 2005. Court-ordered desegregation: successes and failures integrating American schools
since Brown v. Board of Education. J. Hum. Resour. 40(3):559-90.
Reber S. 2010. School desegregation and educational attainment for blacks. J. Hum. Resour.
45(4):893-914.
Rivkin SG. 1994. Residential segregation and school integration. Sociol. of Educ. 67(4):279-92.
Rossell CH. 1976. School desegregation and white flight. Polit. Sci. Quar. 90(4):675-95.
Rusk D. 2002. Trends in school segregation. In Divided We Fail: Coming Together through Public
School Choice, p. 61-87. Wash., DC: The Century Found.
Saatcioglu, A. 2010. Disentangling school- and student-level effects of desegregation and
resegregation on the dropout problem in urban high schools: evidence from the Cleveland
municipal district, 1977-1998. Teach. Coll. Rec.112(5):1391-1442.
Sanbonmatsu L, Kling JR, Duncan GJ, Brooks-Gunn J. 2006. Neighborhoods and academic
achievement: results from the moving to opportunity experiment. J. Hum. Resour. 41(4): 649-691.
Saporito S, Sohoni D. 2007. Mapping educational inequality: concentrations of poverty among poor
and minority students in public schools. Soc. Forces 85(3): 1227-54.
Schwartz H. 2010. Housing policy is school policy: economically integrative housing promotes
40
academic success in Montgomery County, Maryland. Rep. The Century Found.
Stroub KJ, Richards MP. 2013. From resegregation to reintegration: trends in the racial/ethnic
segregation of metropolitan public schools, 1993-2009. Am. Educ. Res. J.
Tatum BD. 1997. Why Are All the Black Kids Sitting Together in the Cafeteria? And Other
Conversations about Race. New York: Basic Books.
Urquiola M. 2005. Does school choice lead to sorting? Evidence from Tiebout variation. Am. Econ.
Rev. 95(4): 1310-26.
U.S. Commission on Civil Rights. 2007. Becoming less separate? School desegregation, justice
department enforcement, and the pursuit of unitary status. Wash, DC: Author.
Vigdor JL. 2011. School desegregation and the black-white test score gap. In Whither Opportunity:
Rising Inequality, Schools, and Children’s Life Chances, ed. GJ Duncan, RJ Murnane, pp. 443-464. New
York: Russell Sage Found.
Vigdor, JL, Ludwig, JO. 2008. Segregation and the test score gap. In Steady Gains and Stalled
Progress: Inequality and the Black-White Test Score Gap, ed. K Magnuson, J Waldfogel. New York;
Russell Sage Found.
Watson T. 2009. Inequality and the measurement of residential segregation by income in American
neighborhoods. Rev. of Income and Wealth 55(3):820-44.
41
Weiner DA, Lutz BF, Ludwig J. 2009. The effects of school desegregation on crime. Work. Pap. No.
15380. Natl. Bur. of Econ. Res.
Welch F, Light A. 1987. New evidence on school desegregation. Wash., DC: United States Commission
on Civ. Rights.
Wells AS, Crain RL. 1994. Perpetuation theory and the long-term effects of school desegregation.
Rev. of Educ. Res. 64:531-55.
Wilson FD. 1985. The impact of school desegregation programs on white public-school enrollment,
1968-1976. Sociol. of Educ. 58(3):137-53.
Yun JT, Reardon SF. 2002. Trends in public school segregation in the South, 1987-2000. Presented
at Resegregation of Southern Schools? Conf., Univ. of North Carolina at Chapel Hill.